Skip to main content
Many production incidents come from third-party dependencies, not your own code. Playwright can simulate these failures so you verify graceful behavior before users see an outage.

​
Failure Modes to Simulate

  • Timeout from external API
  • 5xx response from provider
  • 429 rate limit
  • Invalid payload shape
  • Partial success (one request succeeds, next fails)

​
Route Stubbing Pattern

Use page.route to intercept provider requests and force failure conditions.
tests/resilience/maps-provider.spec.ts
import { test, expect } from '@playwright/test';

test('shows fallback when maps API returns 503', async ({ page }) => {
  await page.route('**/maps-provider/**', async (route) => {
    await route.fulfill({
      status: 503,
      contentType: 'application/json',
      body: JSON.stringify({ error: 'service unavailable' }),
    });
  });

  await page.goto('/locations');
  await expect(page.getByText('Map unavailable')).toBeVisible();
  await expect(page.getByRole('button', { name: 'Retry' })).toBeVisible();
});

​
Timeout Example

test('shows timeout state for slow provider', async ({ page }) => {
  await page.route('**/credit-check/**', async (route) => {
    await new Promise((r) => setTimeout(r, 15000));
    await route.fulfill({ status: 504, body: 'timeout' });
  });

  await page.goto('/apply');
  await expect(page.getByText('Request timed out')).toBeVisible();
});

​
Assertions That Matter

  • User sees actionable message, not generic failure.
  • Retry works and does not duplicate side effects.
  • App remains responsive while waiting.
  • Error is logged with enough metadata for support/debugging.

​
Stably Features to Use for Resilience Monitoring

  • Run resilience suites on Stably Cloud with high parallelism to cover multiple providers quickly.
  • Add Scheduled Test Runs for recurring synthetic outage drills.
  • Configure Alerts & Notifications so third-party regressions are surfaced before customers report them.
  • Use run artifacts in Test Results to inspect traces, screenshots, and failure context when fallback behavior breaks.